This study examines LLM-based AI chatbots as social interaction partners. It identifies young adults’ support-seeking motivations and models their associations with two interaction processes (Perceived Authenticity and Interaction Intensity) and two psychological outcomes (Resilience and AI Dependence). In Study 1, we developed the Motivation Scale for Generative AI (MS-GAI) through theory-driven item generation and interviews and validated it with a survey of Korean young adults (N = 300). Exploratory and confirmatory factor analyses supported an eight-factor, 40-item structure grounded in four social-support domains and supplemented by context-specific motivations. In Study 2, structural equation modeling showed that Emotional Support and Self-Enhancement were positively associated with both Perceived Authenticity and Interaction Intensity. Objective Advice Seeking was positively associated with Perceived Authenticity, whereas Learning Assistance, Daily Problem-Solving, and Entertainment/Mood Diversion were positively associated with Interaction Intensity. Perceived Authenticity was also positively associated with Interaction Intensity. Both interaction processes were positively related to self-reported Resilience and AI Dependence, with Interaction Intensity showing the stronger association with AI Dependence. These findings identify distinct but overlapping interaction pathways through which support-seeking motivations are associated with adaptive and maladaptive outcomes. The study extends understanding of social support in human–AI interaction by providing a validated motivation measure and showing how shared interaction processes relate to both short-term perceived recovery and dependence. The findings are consistent with the Paradoxes of Technology perspective and highlight the need for emotionally supportive yet balanced design and policy safeguards.
In this study, we demonstrate a partially transparent nonvolatile resistive random access memory (ReRAM) device based on Cu2O-graphene core–shell quantum dots (QDs) dispersed in a poly(methyl methacrylate) (PMMA) matrix. The Cu2O-graphene QDs were synthesized through a solvothermal route and exhibited an average diameter of approximately 5 nm. The graphene shell modifies the local interfacial electronic environment and facilitates interfacial charge redistribution around the Cu2O cores. XRD, Raman, FTIR, XPS, and HRTEM analyses collectively support the formation of the Cu2O–graphene hybrid structure, while TEM observations indicate its dispersion within the PMMA matrix. The fabricated devices exhibited clear bipolar resistive switching with an ON/OFF current ratio of approximately 103, stable operation over repeated switching cycles, and experimentally measured retention exceeding 105 s. Extrapolation of the measured retention data suggests that the ON/OFF states would remain distinguishable at 10 years. Detailed current–voltage analysis indicated a transition from ohmic and space-charge-limited conduction to a high-field transport regime consistent with field-assisted tunneling through the PMMA barrier. These results demonstrate that the Cu2O–graphene core–shell QD/PMMA composite is a promising switching material for organic nonvolatile ReRAM and offers a useful platform for transparent and solution-processable next-generation memory systems.
Motivated by the demographic effects, we propose and analyze a novel opinion dynamics model with density-dependent interactions and a nonlinear communication function reminiscent of the celebrated Hegselmann-Krause (H-K) model. The model also incorporates heterogeneity through intrinsic opinion tendencies, leading to richer dynamics. In the homogeneous case, where all agents share the same intrinsic tendency, the relative dissimilarities between opinions converge to zero and a single cluster forms. In contrast, under heterogeneous tendencies, we prove that multiple clusters emerge together with velocity alignment within each cluster, provided suitable structural conditions on the system parameters. These theoretical findings are further supported by numerical simulations, which offer qualitative insights beyond the analytical results.
High-voltage rotating-machines frequently fail due to insulation degradation; however, accelerated life-test (ALT) data are scarce and heavily censored. Moreover, conventional ALT models analyze each voltage class independently, which limits cross-class extrapolation. To address these challenges, this study develops RIPPLE (Rated-voltage-Informed Pseudo-Probabilistic Lifetime Estimation), a Bayesian accelerated failure-time framework that integrates experimental data and specification information to improve lifetime inference. First, censored specimens are transformed into pseudo-labeled lifetimes by learning a mapping from diagnostic time series and design features to failure time using a lightweight Transformer-MLP model ensemble, with predictive uncertainty explicitly quantified. Second, rated-voltage R is incorporated as a design covariate in a LogNormal rated-voltage-informed Temperature-Nonthermal (R-TNT) stress-life model, enabling joint modeling across heterogeneous voltage classes. Third, pseudo-labels are treated as uncertain soft evidence through a truncated latent-variable likelihood, propagating prediction uncertainty into the posterior and enforcing censoring consistency probabilistically. Applied to multi-stress ALT data on 6.6/13.8 kV form-wound coils, RIPPLE reduced the 90% credible-interval width by approximately 90% at 155 degrees C and 2.1 R relative to the censored-likelihood baseline, while substantially improving point-prediction accuracy (MAPE 0.453 vs. 2.213). Simulation studies further support the effectiveness of the proposed framework.
